Authors:
Goloviatinski Sergiy
1
;
Herbelin Ludovic
1
;
José Mancera
2
;
Luis Terán
2
;
Jhonny Pincay
2
;
3
and
Edy Portmann
2
Affiliations:
1
University of Neuchâtel, Avenue du 1er-Mars 26, Neuchâtel, Switzerland
;
2
Human-IST Institute, University of Fribourg, Boulevard de Pérolles 90, Fribourg, Switzerland
;
3
Pontificia Universidad Católica del Ecuador, Av. 12 de Octubre 1076, Quito, Ecuador
Keyword(s):
Graph-based Recommender System, GDPR, Social Networks.
Abstract:
The enforcement of the General Data Protection Regulation (GDPR) in the European Union represents a challenge in designing reliable recommender systems due to user data collection limitations. This work proposes a method to consider GDPR data with a graph-based recommender system to tackle data sparsity and the cold-start problem by representing the data in a knowledge graph. In this work, the authors assess a real dataset provided by Beekeeper AG, a social network company for front-line workers, to model the interactions in a graph database. This work proposes and develops a recommender system on top of the database using the requests made to Beekeeper’s REST API. It explores the API events, neither with knowledge of the content nor the user profiles. Besides, it presents a discussion of multiple approaches for community detection algorithms to retrieve clusters of groups or companies that are part of the social network. This paper proposes several techniques to understand user acti
vity and infer user interactions and events such as likes in posts, comments, and session duration. The recommendation engine presents posts to new and existing users. Thanks to pilot customers who provided consent to access private data, this work verifies the effectiveness of the findings.
(More)